Use glyph indices for font tracking in vector formats

With libraqm, string layout produces glyph indices, not character codes,
and font features may even produce different glyphs for the same
character code (e.g., by picking a different Stylistic Set). Thus we
cannot rely on character codes as unique items within a font, and must
move toward glyph indices everywhere.

Author: QuLogic

lib/matplotlib/_mathtext.py

@@ -38,7 +38,7 @@
 
 if T.TYPE_CHECKING:
     from collections.abc import Iterable
-    from .ft2font import CharacterCodeType, Glyph
+    from .ft2font import CharacterCodeType, Glyph, GlyphIndexType
 
 
 ParserElement.enable_packrat()
@@ -87,7 +87,7 @@ class VectorParse(NamedTuple):
     width: float
     height: float
     depth: float
-    glyphs: list[tuple[FT2Font, float, CharacterCodeType, float, float]]
+    glyphs: list[tuple[FT2Font, float, CharacterCodeType, GlyphIndexType, float, float]]
     rects: list[tuple[float, float, float, float]]
 
 VectorParse.__module__ = "matplotlib.mathtext"
@@ -132,7 +132,8 @@ def __init__(self, box: Box):
     def to_vector(self) -> VectorParse:
         w, h, d = map(
             np.ceil, [self.box.width, self.box.height, self.box.depth])
-        gs = [(info.font, info.fontsize, info.num, ox, h - oy + info.offset)
+        gs = [(info.font, info.fontsize, info.num, info.glyph_index,
+               ox, h - oy + info.offset)
               for ox, oy, info in self.glyphs]
         rs = [(x1, h - y2, x2 - x1, y2 - y1)
               for x1, y1, x2, y2 in self.rects]
@@ -215,6 +216,7 @@ class FontInfo(NamedTuple):
     postscript_name: str
     metrics: FontMetrics
     num: CharacterCodeType
+    glyph_index: GlyphIndexType
     glyph: Glyph
     offset: float
 
@@ -375,7 +377,8 @@ def _get_info(self, fontname: str, font_class: str, sym: str, fontsize: float,
                   dpi: float) -> FontInfo:
         font, num, slanted = self._get_glyph(fontname, font_class, sym)
         font.set_size(fontsize, dpi)
-        glyph = font.load_char(num, flags=self.load_glyph_flags)
+        glyph_index = font.get_char_index(num)
+        glyph = font.load_glyph(glyph_index, flags=self.load_glyph_flags)
 
         xmin, ymin, xmax, ymax = (val / 64 for val in glyph.bbox)
         offset = self._get_offset(font, glyph, fontsize, dpi)
@@ -398,6 +401,7 @@ def _get_info(self, fontname: str, font_class: str, sym: str, fontsize: float,
             postscript_name=font.postscript_name,
             metrics=metrics,
             num=num,
+            glyph_index=glyph_index,
             glyph=glyph,
             offset=offset
         )
@@ -427,8 +431,7 @@ def get_kern(self, font1: str, fontclass1: str, sym1: str, fontsize1: float,
             info1 = self._get_info(font1, fontclass1, sym1, fontsize1, dpi)
             info2 = self._get_info(font2, fontclass2, sym2, fontsize2, dpi)
             font = info1.font
-            return font.get_kerning(font.get_char_index(info1.num),
-                                    font.get_char_index(info2.num),
+            return font.get_kerning(info1.glyph_index, info2.glyph_index,
                                     Kerning.DEFAULT) / 64
         return super().get_kern(font1, fontclass1, sym1, fontsize1,
                                 font2, fontclass2, sym2, fontsize2, dpi)

lib/matplotlib/_text_helpers.py

@@ -14,7 +14,7 @@
 class LayoutItem:
     ft_object: FT2Font
     char: str
-    glyph_idx: GlyphIndexType
+    glyph_index: GlyphIndexType
     x: float
     prev_kern: float
 
@@ -47,19 +47,19 @@ def layout(string, font, *, kern_mode=Kerning.DEFAULT):
     LayoutItem
     """
     x = 0
-    prev_glyph_idx = None
+    prev_glyph_index = None
     char_to_font = font._get_fontmap(string)
     base_font = font
     for char in string:
         # This has done the fallback logic
         font = char_to_font.get(char, base_font)
-        glyph_idx = font.get_char_index(ord(char))
+        glyph_index = font.get_char_index(ord(char))
         kern = (
-            base_font.get_kerning(prev_glyph_idx, glyph_idx, kern_mode) / 64
-            if prev_glyph_idx is not None else 0.
+            base_font.get_kerning(prev_glyph_index, glyph_index, kern_mode) / 64
+            if prev_glyph_index is not None else 0.
         )
         x += kern
-        glyph = font.load_glyph(glyph_idx, flags=LoadFlags.NO_HINTING)
-        yield LayoutItem(font, char, glyph_idx, x, kern)
+        glyph = font.load_glyph(glyph_index, flags=LoadFlags.NO_HINTING)
+        yield LayoutItem(font, char, glyph_index, x, kern)
         x += glyph.linearHoriAdvance / 65536
-        prev_glyph_idx = glyph_idx
+        prev_glyph_index = glyph_index

lib/matplotlib/backends/_backend_pdf_ps.py

@@ -2,9 +2,12 @@
 Common functionality between the PDF and PS backends.
 """
 
+from __future__ import annotations
+
 from io import BytesIO
 import functools
 import logging
+import typing
 
 from fontTools import subset
 
@@ -14,33 +17,38 @@
 from ..backend_bases import RendererBase
 
 
+if typing.TYPE_CHECKING:
+    from .ft2font import FT2Font, GlyphIndexType
+    from fontTools.ttLib import TTFont
+
+
 @functools.lru_cache(50)
 def _cached_get_afm_from_fname(fname):
     with open(fname, "rb") as fh:
         return AFM(fh)
 
 
-def get_glyphs_subset(fontfile, characters):
+def get_glyphs_subset(fontfile: str, glyphs: set[GlyphIndexType]) -> TTFont:
     """
-    Subset a TTF font
+    Subset a TTF font.
 
-    Reads the named fontfile and restricts the font to the characters.
+    Reads the named fontfile and restricts the font to the glyphs.
 
     Parameters
     ----------
     fontfile : str
         Path to the font file
-    characters : str
-        Continuous set of characters to include in subset
+    glyphs : set[GlyphIndexType]
+        Set of glyph indices to include in subset.
 
     Returns
     -------
     fontTools.ttLib.ttFont.TTFont
         An open font object representing the subset, which needs to
         be closed by the caller.
     """
-
-    options = subset.Options(glyph_names=True, recommended_glyphs=True)
+    options = subset.Options(glyph_names=True, recommended_glyphs=True,
+                             retain_gids=True)
 
     # Prevent subsetting extra tables.
     options.drop_tables += [
@@ -71,7 +79,7 @@ def get_glyphs_subset(fontfile, characters):
 
     font = subset.load_font(fontfile, options)
     subsetter = subset.Subsetter(options=options)
-    subsetter.populate(text=characters)
+    subsetter.populate(gids=glyphs)
     subsetter.subset(font)
     return font
 
@@ -97,23 +105,24 @@ def font_as_file(font):
 
 class CharacterTracker:
     """
-    Helper for font subsetting by the pdf and ps backends.
+    Helper for font subsetting by the PDF and PS backends.
 
-    Maintains a mapping of font paths to the set of character codepoints that
-    are being used from that font.
+    Maintains a mapping of font paths to the set of glyphs that are being used from that
+    font.
     """
 
-    def __init__(self):
-        self.used = {}
+    def __init__(self) -> None:
+        self.used: dict[str, set[GlyphIndexType]] = {}
 
-    def track(self, font, s):
+    def track(self, font: FT2Font, s: str) -> None:
         """Record that string *s* is being typeset using font *font*."""
         char_to_font = font._get_fontmap(s)
         for _c, _f in char_to_font.items():
-            self.used.setdefault(_f.fname, set()).add(ord(_c))
+            glyph_index = _f.get_char_index(ord(_c))
+            self.used.setdefault(_f.fname, set()).add(glyph_index)
 
-    def track_glyph(self, font, glyph):
-        """Record that codepoint *glyph* is being typeset using font *font*."""
+    def track_glyph(self, font: FT2Font, glyph: GlyphIndexType) -> None:
+        """Record that glyph index *glyph* is being typeset using font *font*."""
         self.used.setdefault(font.fname, set()).add(glyph)
 
 

lib/matplotlib/backends/backend_cairo.py

@@ -8,6 +8,7 @@
 
 import functools
 import gzip
+import itertools
 import math
 
 import numpy as np
@@ -248,13 +249,12 @@ def _draw_mathtext(self, gc, x, y, s, prop, angle):
         if angle:
             ctx.rotate(np.deg2rad(-angle))
 
-        for font, fontsize, idx, ox, oy in glyphs:
+        for (font, fontsize), font_glyphs in itertools.groupby(
+                glyphs, key=lambda info: (info[0], info[1])):
             ctx.new_path()
-            ctx.move_to(ox, -oy)
-            ctx.select_font_face(
-                *_cairo_font_args_from_font_prop(ttfFontProperty(font)))
+            ctx.select_font_face(*_cairo_font_args_from_font_prop(ttfFontProperty(font)))
             ctx.set_font_size(self.points_to_pixels(fontsize))
-            ctx.show_text(chr(idx))
+            ctx.show_glyphs([(idx, ox, -oy) for _, _, idx, ox, oy in font_glyphs])
 
         for ox, oy, w, h in rects:
             ctx.new_path()